Extended Data Fig. 2: SRD results for total (bottom row) and average (top row) misinformation tweets and retweets, for deplatformed and not-deplatformed users.
From: Post-January 6th deplatforming reduced the reach of misinformation on Twitter
Sample size includes 546 observations (days) on average across groups (x-axis), 404 before and 136 after. The effective number of observations is 64.31 days before and after on average. The estimation excludes data between Jan 6 (cutoff point) and 12 (included). January 6th is the score value 0, and January 12th the score value 1. Optimal bandwidth of 32.6 days with triangular kernel and order-one polynomial. Bars indicate 95% robust bias-corrected confidence intervals.
